lable at ScienceDirect

Vacuum xxx (2016) 1e5

Contents lists avai

Vacuum

journal homepage: www.elsevier .com/locate/vacuum

Etching of latent tracks in amorphous SiO2 and Si3N4: Simulation andexperiment

L. Vlasukova a, *, F. Komarov b, V. Yuvchenko b, L. Baran a, O. Milchanin b,А. Dauletbekova c, A. Alzhanova c, A. Akilbekov c

a Belarusian State University, 4 F. Skorina Ave., 220030 Minsk, Republic of Belarusb A. N. Sevchenko Institute of Applied Physical Problems of Belarusian State University, 7 Kurchatova Str., 220045 Minsk, Republic of Belarusc Faculty of Physics and Technical Sciences, L. N. Gumilyov Eurasian National University, 5Munaitpasov Str., 010008 Astana, Republic of Kazakhstan

a r t i c l e i n f o

Article history:Received 23 October 2015Received in revised form30 December 2015Accepted 31 December 2015Available online xxx

Keywords:SiO2/SiSi3N4/SiSwift heavy ionThermal spike modelLatent track etchingNanopores

* Corresponding author. Belarusian State Universi220045, Belarus.

E-mail address: vlasukova@bsu.by (L. Vlasukova).

http://dx.doi.org/10.1016/j.vacuum.2015.12.0230042-207X/© 2016 Elsevier Ltd. All rights reserved.

Please cite this article in press as: L. Vlasukov(2016), http://dx.doi.org/10.1016/j.vacuum.2

a b s t r a c t

The latent track formation in silicon dioxide (SiO2) and silicon nitride (Si3N4) irradiated with swift heavyions (SHI) has been studied using computer simulation in the frame of the thermal spike model. We havecalculated radii and lifetime of the molten regions, or the regions heated to the melting point, formed inSiO2 and Si3N4 along the ion trajectories for F, S, Cl, Ar, Kr and Xe ions in an energy range of (28e200)MeV. The radius of the molten region was chosen as a criterion for track “etchability” in the case of SiO2.The results of computer simulation have been compared with the experimental results of track etching in4% aqueous solution of hydrofluoric acid (HF). The validity of the criterion taken for the creation ofhomogeneous “etchable” ion tracks in SiO2, namely, the formation of a molten region with radius largerthan 3.0 nm [1] in the matrix along the ion trajectory, has been confirmed. It has been found that boththe etched track depth and diameter increase with the fluence for the same type of ion species. Underour experimental conditions, for the case of Si3N4 we failed to etch regular conical channels with uniformsize distribution.

© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

When SHI penetrates dielectric material, it induces a damagedregion called “latent track”. This damaged region often exhibits adifferent chemical reactivity than unmodified surrounding. Thechemical treatment of latent tracks in appropriate etchants canresult in the formation of nanochannels in the dielectric matrix[1e3]. This approach is widely used to manufacture track mem-branes in polymer films. Such membranes are applied as fine filtersin organic synthesis, bioengineering and medicine. More recentlyan attention has been given to track etching of thin SiO2 and Si3N4layers grown on Si wafers because of possible applications in Sitechnology. Nanoporous silicon-based insulators grown on Si wa-fers can be used as a template to produce arrays of nanoclusters andnanotubes for nanoelectronic devices [4], as active elements ofbiosensors [5,6] as well as in new generation of track membraneswith improved thermal and chemical resistance.

ty, Kurchatova Str. 5, Minsk

a, et al., Etching of latent trac015.12.023

It is important to evaluate the crucial factors of track etching forreproducible creation of dielectric layers with high nanochanneldensity (up to 1011 cm�2). Among these factors the most importantones are: probability of formation of the latent track around eachincident ion and size of pore, which may form in the place of alatent track during etchant treatment. A commonly used track“etchability” criterion is the threshold electronic stopping powerSe0. According to different data Se0 varies from 1.5 to 4 keV/nm forSiO2 [7e9]. The other criterion for “etchability” of tracks was sug-gested by Dallanora et al. [1]. The effects of the swift heavy ionpassage through SiO2 and some other insulators are well describedusing the thermal spike model [10]. The model involves the ther-malization of the electronic subsystem of a solid within a time in-terval of about 10�14 s. A few picoseconds later, theelectronephonon interaction leads to fast heating of the regionalong the fast ion trajectory. The process of energy transfer from theelectronic to atomic subsystem of a solid is described by a system oftwo differential equations. The model involves one fitting param-eter: the electronephonon interaction mean free path l. If thedensity of the energy released in electronic excitations is suffi-ciently high, we observe the melting of the material and the

ks in amorphous SiO2 and Si3N4: Simulation and experiment, Vacuum

L. Vlasukova et al. / Vacuum xxx (2016) 1e52

formation of a cylindrical domain with a diameter of a few nano-meters - the future latent track. A few tens of picoseconds later, themelt cools to the temperature of the surrounding matrix. Accordingto the thermal spike model the etchable track results from a rapidquenching of a molten liquid phase. The authors of [1] have re-ported that the track etching appears if the molten region radius isat least 1.6 nm. Homogeneous etching comes out only for latenttracks with radii larger than 3.0 nm. In a recent paper [11] wecompared the results of computer simulation of track formationand experimental results for track etching in amorphous SiO2irradiated with swift heavy ions. We found a good correlation be-tween track “etchability” and threshold melt region radius for thebeginning of track etching and for homogeneous track etchingpredicted in Ref. [1].

In this study, we have estimated the possibility of formation ofthe “etchable” tracks in Si-based insulators using swift ion irradi-ation regimes available in an ion cyclotron accelerator DC-60(Astana, Kazakhstan). We have also reported the experimental re-sults on the latent track etching in SiO2 and Si3N4 layers on Siwafers.

2. Experimental

Our simulation is based on the thermal spike model using asoftware systemwe developed. The details are described elsewhere[3]. We calculated the latent track parameters in SiO2 and Si3N4 fora number of ion species and energies taking into account the ionaccelerator DC-60 possibilities. For SiO2 the thermophysical pa-rameters needed for the calculations have been taken fromRefs. [1],the parameter l was 4 nm. For Si3N4 the parameter l was 4.8 nm[10]. We failed to find reference data on latent heat of melting ofthis latter material. In addition, sublimation or thermal decompo-sition of the material can take place upon the heating of Si3N4,along with solid phase melting. As a track region in Si3N4 we tookthe region heated to themelting point instead of the molten region.The latent heat of melting of Si3N4 was taken to be 1 � 108 J kg�1,which is higher by two orders of magnitude than the average valuesfor semiconductors. The electronic stopping power Se was calcu-lated by means of SRIM 2008 [12]. For the electronic stopping po-wer calculation, the densities of SiO2 and Si3N4 were taken to be 2.2and 2.85 g/cm3, respectively [13]. We have calculated the radii andlifetimes of the molten regions along the swift ion trajectories. Themaximum range of d-electrons emitted during the ion passagethrough the dielectric matrix was calculated according to the

Table 1The latent tracks parameters in SiO2 calculated for group I.

Ion type and energy, MeV The molten region ra

F (28) e

S (47) 1.6Cl (43) 1.9Ar (38) 2.2Ar (54) 2.0

Table 2The latent tracks parameters in SiO2 calculated for ion species from group II.

Ion type, energy, MeV Se, keV/nm (SRIM 2008) The molten region radius r1, nm

Kr (58) 8.61 4.4Kr (86) 9.23 4.5Kr (140) 9.69 4.4Xe (65.5) 10.66 5.4Xe (132) 13.51 5.9Xe (200) 14.96 6.1

Please cite this article in press as: L. Vlasukova, et al., Etching of latent trac(2016), http://dx.doi.org/10.1016/j.vacuum.2015.12.023

method in Ref. [14].SiO2 and Si3N4 films were grown on n-type (100)-Si substrates

by plasma-enhanced chemical vapour deposition at 300 �C. Thethicknesses of films were ~1000 and ~600 nm for SiO2 and Si3N4respectively. The samples of size 1 � 1 cm2 were cut from SiO2/Siand Si3N4/Si wafers and irradiated with 38 MeV-Ar, 58 MeV-Kr and200 MeV-Xe ions in the fluence range of (2 � 10 8 - 1 � 10 12) cm�2

at normal incidence at the ion cyclotron accelerator DC-60 (Astana,Kazakhstan). The irradiated samples were treated in a 4% hydro-fluoric acid (HF) aqueous solution at room temperature for 6 min inthe case of SiO2/Si and for 40 min in the case of Si3N4/Si. In order toavoid artefacts all SiO2/Si samples were etched in the same etchprocess. The same procedure was carried out for Si3N4/Si samples.Then the surface of the etched samples was investigated using thescanning electron microscope JSM 7500F and the atomic forcemicroscope Solver P-47 in the tapping mode. Mean values of theetched pore diameter D averaged over 3e5 images for each specificirradiation of SiO2 matrix were obtained from SEM images. Meanvalues of the pore depth z averaged over 10e15 tracks were esti-mated from AFM profiles. It should be noted that pore depthsdetermined from AFM images may be distorted by the finite size ofthe probe tip. However, one can use AFM data for preliminaryestimation of the pore parameters.

3. Results

3.1. Simulation of latent track formation in SiO2 and Si3N4

Ion species chosen for simulation can be divided in two groups.The first one, group I, includes comparatively “light” ions withenergies of tens of MeV, namely, 19F (28 MeV), 32S (47 MeV), 35Cl(43MeV), and 40Ar (38, 54MeV). The group II consists of heavy 84Krand 132Xe. In order to estimate the ion velocity effect [15] on latenttrack parameters we have simulated the different energies of pro-jectiles, namely, 56, 86 and 140 MeV for 84Kr and 65.5, 132 and200 MeV for 132Xe and analyzed the maximum range of d-electronsfor these projectiles, additionally.

Table 1 presents the latent tracks parameters in SiO2 calculatedfor ions of group I. One can see that the melting point is not ach-ieved for 19F with an energy of 28 MeV. A formation of moltenregions along ion trajectories takes place for other species. It is aproblem to expect that each ion impact should result in an etchedtrack formation because melt region radii are lower than thresholdvalue for homogeneous track etching (3.0 nm). Table 2 presents the

dius r, nm The molten region lifetime t, ps

e

1.41.92.72.2

The molten region lifetime t1, ps The maximum range of d-electrons rm, nm

11.3 53.5811.5 102.8411.1 232.0817.6 30.7521.5 99.2123.0 198.79

ks in amorphous SiO2 and Si3N4: Simulation and experiment, Vacuum

L. Vlasukova et al. / Vacuum xxx (2016) 1e5 3

results of computer simulation of latent track formation in SiO2 forion species of group II. All calculated molten region radii exceed3.0 nm in the case of SiO2 irradiated with Kr and Xe ions, and wecan expect that each swift ion creates latent track. Therefore, it ispossible to achieve a reproducible fabrication of nanochannels withregular shape in SiO2 layers as well as to govern the nanochanneldensity by means of ion fluence.

It should be noted that the latent track radius and lifetime do notincrease with the ion energy in the case of the both materialsirradiated with Kr ions. At the same time the electronic stoppingpower is increasing substantially. That is why, we should considerthe “velocity effect” when evaluating the efficiency of latent track

Fig. 1. Scanning electron microscope images of the etched tracks in SiO2 irradiatedwith 38 MeV-Ar ions to fluence of 1 � 10 9 (A) and 1 � 10 11 cm�2 (B) and etched in 4%HF for 6 min. White scale bar is 100 nm.

Table 3The latent tracks parameters in Si3N4 calculated for ion species from group II.

Ion type,energy, MeV

The electronic stopping powerSe, keV/nm

The radius of the region heated to themelting point r2, nm

Kr (58) 11.06 2.9Kr (86) 11.92 2.0Kr (140) 12.38 2.8Xe (65.5) 13.55 3.9Xe (132) 17.45 4.5Xe (200) 19.18 4.7

Please cite this article in press as: L. Vlasukova, et al., Etching of latent trac(2016), http://dx.doi.org/10.1016/j.vacuum.2015.12.023

formation [15]. During the ion's passage through the target, high-energy d-electrons are emitted. With the similar Se, most of theenergy released during the passage of a slower ion is localized in anarrow region along the track. In the case of a faster ion, most of thereleased energy is carried out away by d-electrons. Thus, the slowerion causes greater heating of the region of a future track.

The results of computer simulation of latent track formation inSi3N4 for of group II species are shown in Table 3. One can see that aregion heated to the melting point is formed during the irradiationwith all discussed ion species in the case of Si3N4. It should be notedthat Se is higher in Si3N4 irradiated with Kr and Xe ions in com-parison with SiO2 irradiated with the same ion species in the same

Fig. 2. Scanning electron microscope images of the etched tracks in Si3N4 irradiatedwith 200 MeV-Xe ions to fluence of 2 � 109 (A) and 2 � 1010 cm�2 (B) and etched in 4%HF for 40 min. White scale bar is 1 mm.

The lifetime of the region heated to themelting point t2, ps

The maximum range of d-electrons rm, nm

1.2 42.31.3 81.191.2 183.222.4 24.793.3 78.333.6 156.94

ks in amorphous SiO2 and Si3N4: Simulation and experiment, Vacuum

L. Vlasukova et al. / Vacuum xxx (2016) 1e54

energy range (see Tables 2 and 3). However, the calculated radii ofthe region heated to the melting point are substantially smaller incomparison with the radii of the molten region for SiO2. Also, thecalculated lifetimes of the region heated to the melting point areshorter in comparison with the lifetimes of the molten region forSiO2. One can conclude that the formation of the region heated tothe melting point in Si3N4 requires more energy expenditure incomparison with the formation of the molten region in SiO2.

3.2. Latent track etching

Fig. 1 shows the SEM images of etched tracks in SiO2 irradiatedwith 38 MeV-Ar ions. One can see that pore diameters vary from~20 to 100 nm. The track etching efficiency, defined as x ¼ Np/F,where Np is the number of pores per unit area and F is a fluence,did not exceed 70% (Fig. 1A). This corresponds perfectly withtheoretical predictions. The melt region radius in this case is 2.2 nm(Table 1). It is larger than the threshold value for initiation of tracketching (1.6 nm) and lower than the threshold value for homoge-neous track etching.Wewould not expect that the pore density willbe equal to fluence, and that the formation of pores with the same

Fig. 3. AFM images and profiles of the etched SiO2 surfaces: the samples irradiated with 38(E, F, I, J), 1 � 109 (A, B), 2 � 10 9 (K, L), 1 � 1010 (C, D) and 2 � 1010 cm �2 (G, H).

Table 4Irradiation regimes, mean depths and sizes of the etched pores in the track regions of Si

Ion type and energy Ion fluence, cm�2 Mean pore dept

Ar (38 MeV) 1 � 109 9.5 ± 6.11 � 1010 19.4 ± 4.9

Kr (58 MeV) 2 � 108 68.4 ± 8.12 � 109 125.2 ± 16.1

Xe (200 MeV) 2 � 108 92.2 ± 21.02 � 109 115.2 ± 36.5

Please cite this article in press as: L. Vlasukova, et al., Etching of latent trac(2016), http://dx.doi.org/10.1016/j.vacuum.2015.12.023

size will take place.For SiO2 irradiated with 58 MeV-Kr ions to 2 � 10 8 and 2 � 10

9 cm�2, the pore diameter is nearly the same, and the pore densitycorresponds to fluence (data not shown). The melt region radius inthis case is 4.4 nm (Table 2), and homogeneous track etching takesplace. The fluence growth to 2 � 10 10 cm �2 and higher results inoverlapping of etched tracks. The same picture is observed in thecase of the 200 MeV-Xe irradiation to fluences of 2 � 10 8 and2 � 10 9 cm�2, namely, the etching results in the formation of asystem of nanochannels with regular form in the SiO2 layer. Thenanochannel density corresponds to fluence in this fluence range,too. For higher fluence etched track overlapping is observed.

Fig. 2 demonstrates the etched tracks in Si3N4 irradiated with200MeV-Xe ions. One can see shallow pits with irregular shape anddifferent size. These structures are hard to estimate quantitativelybecause their features almost merge with the asperities of thesurface roughness. A rough estimation yields a pore densityz3.5 � 10 8 cm �2 for the sample irradiated with a fluence of2 � 109 cm �2. It corresponds to a track etching efficiency of ~18%.These results are in accordance with our data for Si3N4 irradiatedwith 180 MeVeW ions [16]. Then we observed a formation of

MeV Ar (A, B, C, D), 56 MeV Kr (E, F, G, H), 200 MeV Xe (I, J, K, L) with fluence of 2 � 108

O2.

h z, nm (AFM data) Mean etched pore diameter D, nm (SEM data)

e

e

123.8 ± 6.5137.6 ± 14.5123.6 ± 9.0145.6 ± 9.1

ks in amorphous SiO2 and Si3N4: Simulation and experiment, Vacuum

L. Vlasukova et al. / Vacuum xxx (2016) 1e5 5

discontinuous tracks and recorded a low track etching efficiency,too. For 180MeVeW ions the calculated radius of the region heatedto themelting point was 5.3 nm and its lifetimewas 4.7 ps. For 200-MeV Xe ions these values are 4.7 nm and 3.6 ps respectively.We canassume that the threshold for homogeneous track etching of siliconnitride is not achieved in our experiment.

Fig. 3 shows the AFM images and profiles of the etched SiO2surfaces irradiated with 38 MeV-Ar, 58 MeV-Kr and 200 MeV-Xeions in the range of fluences of (2e20) � 108 cm�2. In the case ofirradiationwith Ar up to 10 9 cm�2 the depths of etched tracks differsubstantially (Fig. 3B). It should be noted that etched track over-lapping was not observed up to an Ar ion fluence of 1010 cm�2

(Fig. 3D) but the fluence growth to 10 11 cm�2 results in etched trackoverlapping. For the irradiation with 58 MeV-Kr and 200 MeV-Xeions track overlapping is already observed at a fluence of2 � 1010 cm�2. The example of track overlapping is shown in Fig. 3G, H for SiO2 irradiated with 58 MeV-Kr. For low fluences(2e20) � 108 cm�2, in the absence of track overlapping, it ispossible to estimate the etched track depth and size. The meanvalues of these parameters calculated from AFM and SEM imagesare summarized in Table 4. (In the case of Ar irradiation, etchedpore diameters varied from ~20 to 100 nm, and the calculation of Dwas not carried out). Analysis of the Tables 1, 2 and 4 shows atendency of increasing of etched track depth and diameter with theradius and lifetime of the calculated molten region. One can seefrom Table 4 that the etched track depth and diameter increaseswith the fluence for the same type of ion species. This effect is asubject of future investigation. Though, one can suppose thatradiation-induced damage accumulation results in the growth ofchemical activity of track regions as well as in the whole SiO2matrix.

4. Conclusions

On the basis of the thermal spike model we have calculated radiiand lifetimes of the molten regions, or the regions heated to themelting point, formed in amorphous SiO2 and Si3N4 irradiated withswift ions. The radius of the molten region along a swift ion tra-jectory was chosen as a criterion for track “etchability” in the caseof SiO2.

It was shown that comparatively “light” ions with energies ofsome tens of MeV, namely, 32S (47 MeV), 35Cl (43 MeV), and 40Ar(38, 54 MeV) can produce latent tracks. However it is unrealistic toexpect that each ion impact should result in the formation of etchedtrack because the calculated molten region radii are smaller thanthe threshold melt region radius for a homogeneous track etching.In the case of SiO2 irradiation with 84Kr (56, 86 and 140 MeV) and132Xe (65.5, 132 and 200 MeV), it is possible to achieve a repro-ducible fabrication of regular conical nanochannels with uniformsize distribution in SiO2 layers. It was shown that one shouldconsider the “velocity effect” when evaluating the efficiency oflatent track formation.

Si3N4 track formation process probably requires more energy

Please cite this article in press as: L. Vlasukova, et al., Etching of latent trac(2016), http://dx.doi.org/10.1016/j.vacuum.2015.12.023

expenditure in comparison with SiO2. The values of Se in Si3N4irradiated with Kr and Xe ions are higher in comparison with thosefor SiO2 irradiated with the same ion species in the same energyrange. The calculated radii and lifetimes of the region heated to themelting point are substantially smaller in comparisonwith the radiiand lifetimes of the molten region for SiO2.

The simulation results on latent track formation in SiO2 andSi3N4 have been compared with the experimental etching data ofSiO2 and Si3N4 irradiated with 38 MeV-Ar, 58 MeV-Kr and200 MeV-Xe ions in the fluence range of (2 � 108 - 1 � 1012) cm�2.The validity of the criterion taken for the creation of homogeneousetchable ion tracks in SiO2, namely, the formation of a molten re-gionwith radius larger than 3.0 nm [1] in thematrix along the swiftion trajectory, has been confirmed. It has been found that theetched track depth and diameter increase with the fluence for thesame type of ion species. One can suppose that radiation-induceddamage accumulation results in the growth of chemical activityof track regions as well as that of the SiO2 matrix as a whole. In thecase of Si3N4 the low etching efficiency and the considerable dif-ference in the diameters of the etched pores allow us to assumethat there is the formation of discontinuous tracks.

Acknowledgements

The studywas supported by the State Committee on Science andTechnology of the Republic of Belarus (grant No.T14KAZ-002) andby the Ministry of Education and Science of the Republic ofKazakhstan (2281/GF4).

References

[1] A. Dallanora, D.A. Marcondes, T.L. Bermudez, G.G. Fichtner, C. Trautmann,M. Toulemonde, R.M. Papaleo, J. Appl. Phys. 104 (8) (2008) 024307.

[2] F. Bergamini, M. Bianconi, S. Cristiani, L. Gallerani, A. Nubile, S. Petrini,S. Sugliani, Nucl. Instr. Meth. B 266 (2008) 2475e2480.

[3] L.A. Vlasukova, F.F. Komarov, V.N. Yuvchenko, O.V. Mil’chanin, A.Yu Didyk,V.A. Skuratov, S.B. Kislitsyn, Bull. Russ. Acad. Sci. Phys. 76 (2012) 582e587.

[4] K. Hoppe, W.R. Fahrner, D. Fink, et al., Nucl. Instr. Methods, B 266 (2008) 1642.[5] M. Fujimaki, C. Rocksthul, X. Wang, et al., Opt. Exp. 16 (2008) 64068.[6] N. Ferting, R.H. Blick, J.C. Berhends, Biophys. J. 18 (2002) 3056.[7] R.L. Fleischer (Ed.), Nuclear Tracks in Solids, Univ. California Press, Berkeley,

1975, pp. 23e27.[8] A. Sigrist, R. Balzer, Helv. Phys. Acta. 50 (1977) 49.[9] J. Jensen, M. Skupinski, A. Razpet, G. Possnert, Nucl. Instr. Meth. B 245 (2000)

269e273.[10] M. Toulemonde, C. Dufour, A. Meftah, E. Paumier, Nucl. Instr. Meth. B 166e167

(2000) 903e912.[11] L.A. Vlasukova, F.F. Komarov, V.N. Yuvchenko, W. Wesch, E. Wendler,

A.Yu Didyk, V.A. Skuratov, S.B. Kislitsin, Vacuum 105 (2014) 107e110.[12] J.F. Ziegler, J.P. Biersack, U. Littmark, Stopping Power and Ranges of Ions in

Matter. Vol. I, Pergamon Press, New York, 1985.[13] Gmelin Handbook of Inorganic and Organometallic Chemistry. Si. Suppl. Vol.

B5e. eighth ed.. Springer-Verlag. 1994. P. 264.[14] M.P. Waligorski, R.N. Hamm, R. Katz, Int. J. Radiat. Appl. Instrum. Part D. 11

(1986) 309e319.[15] A. Meftah, F. Brisard, J.M. Costantini, M. Hage-Ali, J.P. Stoquert, F. Studer,

M. Toulemonde, Phys. Rev. B 49 (1994) 12457e12463.[16] L.A. Vlasukova, F.F. Komarov, V.N. Yuvchenko, V.A. Skuratov, A. Yu Didyk,

D.V. Plyakin, Bull. Russ. Acad. Sci. Phys. 74 (2010) 206e208.

ks in amorphous SiO2 and Si3N4: Simulation and experiment, Vacuum